Rare and Threatened Pandanaceae Indicate Sites in Need of Protection

Oryx Vol 41 No 2 April 2007

Identification of priority areas for plant conservation in Madagascar using Red List criteria: rare and threatened Pandanaceae indicate sites in need of protection

Martin W. Callmander, George E. Schatz, Porter P. Lowry II, Michel O. Laivao, Jeannie Raharimampionona, Sylvie Andriambololonera, Tantely Raminosoa and Trisha K. Consiglio

Abstract A major problem in establishing effective cell sizes to estimate area of occupancy. Using a grid cell protocols for conserving Madagascar’s biodiversity is size within the range suggested by IUCN overestimates the lack of reliable information for the identification of threatened status if based solely upon specimen data. priority sites in need of protection. Analyses of field data For poorly inventoried countries such as Madagascar and information from herbarium collections for mem- measures of range size based on such data should be bers of the plant family Pandanaceae (85 spp. of complemented with field observations to determine Pandanus; 6 spp. of Martellidendron) showed how risk population size, sensitivity to disturbance, and specific of extinction assessments can inform conservation threats to habitat and therefore potential population planning. Application of IUCN Red List categories and decline. The analysis of such data can make an criteria showed that 91% of the species are threatened. important contribution to the conservation planning Mapping occurrence revealed centres of richness and process by identifying threatened species and revealing rarity as well as gaps in Madagascar’s existing protected the highest priority sites for their conservation. area network. Protection of 10 additional sites would be required to encompass the 19 species currently lacking Keywords IUCN Red List, Madagascar, Pandanaceae, representation in the reserve network, within which east protected areas, threatened species. coast littoral forests are particularly under represented and important. The effect of scale on assessments of risk This paper contains supplementary material that can of extinction was explored by applying different grid only be found online at http://journals.cambridge.org

Introduction for conservation and for the study of evolutionary processes (Myers & Knoll, 2001). The Malagasy flora is characterized by high levels of At the 2003 World Parks Congress in Durban, South diversity (12,000–14,000 species; Goodman & Benstead, Africa, the Malagasy President, Marc Ravalomanana, 2005; Phillipson et al., 2006) and endemism (.90%; announced plans to triple the size of the protected areas Schatz, 2001), and is threatened by unsustainable land network by 2008, from 1.7 to 6 million ha (nearly 10% of use practices, such as shifting cultivation and fire, that the country). This goal has been integrated into Phase 3 have left ,10% of the original vegetation intact (Myers of the National Environmental Action Plan, and an et al., 2000). Madagascar is regarded as a high priority initial set of target areas has been identified for inclusion in an expanded protected area network. However, research and analysis is needed to ensure maximal Martin W. Callmander (Corresponding author) Missouri Botanical Garden, P.O. Box 299, St. Louis, MO 63166-0299, USA, and University of Neuchaˆtel, inclusion of the country’s biodiversity. Application of Laboratoire de Botanique Evolutive, Case Postale 158, 2009 Neuchaˆtel, the IUCN Red List criteria, a widely recognized tool for Switzerland. E-mail [email protected] identifying threatened species, offers a powerful method George E. Schatz, Porter P. Lowry II* and Trisha K. Consiglio Missouri to identify priority sites for protection by providing Botanical Garden, P.O. Box 299 St. Louis, MO 63166-0299, USA. information on the conservation status of species in the

Michel O. Laivao, Jeannie Raharimampionona, Sylvie Andriambololonera wild (Rodrigues et al., 2006). In Madagascar, however, and Tantely Raminosoa Madagascar Research and Conservation Program, only 425 plant species have been assessed to date, of Missouri Botanical Garden, BP 3391 Antananarivo 101, Madagascar. which only 210 were evaluated using the current criteria *Also at: De´partement Syste´matique et Evolution, Muse´um National d’Histoire (IUCN, 2006). Many of the species that have not been Naturelle, Case Postale 39, 57 rue Cuvier, 75231 Paris CEDEX 05, France. assessed are so poorly known or rare that up-to-date Received 4 October 2005. Revision requested 24 April 2006. distribution information is unavailable. To address this Accepted 13 February 2007. we have developed an efficient, robust procedure for

168 ß 2007 FFI, Oryx, 41(2), 168–176 doi:10.1017/S0030605307001731 Printed in the United Kingdom

Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 13:48:47, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605307001731 Malagasy Pandanaceae 169

evaluating restricted range plant species using primary underestimating the number of threatened species; this occurrence data from herbarium collections and recent approach has been followed here. We used the four Red field observations ( Schatz, 2002; Raharimampionona et al., List criteria and subcriteria (IUCN, 2001) to determine 2006). This procedure also identifies sites with high threat: rapid population decline (Criterion A), limited species richness and concentrations of rare species that geographic range and small population size linked with constitute prospective conservation priorities. Here we fragmentation, decline or fluctuation (Criteria B and C), provide an example of the application of these comple- and very small population size (Criterion D); we did not mentary approaches using the plant family Pandanaceae. attempt to analyse extinction risk quantitatively using Pandanaceae comprise four genera of arborescent or Criterion E. lianoid dioecious monocotyledons, two of which occur For each species of Pandanaceae the geographical in Madagascar, Martellidendron (7 spp., all but one information system Arcview v. 3.3 (ESRI, Redlands, USA) restricted to the island) and the widespread was used to measure three parameters required for Paleotropical genus Pandanus (c. 600 spp., with 85 spp. categorization using the Red List criteria (IUCN, 2001): endemic to Madagascar). The taxonomy of both genera extent of occurrence (EOO, minimum convex polygon has been revised recently (Laivao et al., 2000, 2006; containing all points of occurrence), area of occupancy Callmander et al., 2001, 2003a,b,c; Callmander & Laivao, (AOO, area estimated by superimposing a grid onto 2002) and, combined with information from recent field occurrence points and calculating the cumulative area of work, provides data suitable for illustrating our proce- cells occupied by a species), and predicted future dures for identifying conservation priorities. Species of decline (PFD, estimated continuing reduction in a Pandanaceae grow in all tropical forest types (dry to species’ AOO). For species restricted to primary vegeta- humid) and exhibit both widespread and restricted tion PFD was estimated using (AOO outside protected distributions. By applying IUCN Red List categories, areas / total AOO) * 100, which is based on two calculating richness and rarity, and identifying areas assumptions that reasonably reflect the current situation with high diversity of endemic Malagasy Pandanaceae, for long-lived arborescent plants in Madagascar (Schatz we (1) illustrate how assessments using herbarium data et al., 2000; Good et al., 2006): within a few decades (3 can inform conservation planning, (2) show how the generations for most Pandanaceae, estimated to be 30–50 assessments can be used to evaluate the effectiveness of years, depending on the species) ongoing habitat the existing protected area network and identify high degradation and destruction will leave few significant priority sites for establishing new protected areas and/ areas of primary vegetation outside the protected area or implementing alternative conservation strategies, (3) network, and loss of primary vegetation within pro- explore difficulties encountered in applying the Red List tected areas will be minimal. While some unprotected criteria relating to species distributions, and (4) illustrate areas may escape destruction (especially on steep or some problems involved in using herbarium collections rugged terrain), it is nevertheless unlikely that many and field observations for Red List assessments and populations of long-lived plants will survive outside show how occurrence data can be analysed most protected areas after 3 generations, given Madagascar’s effectively to provide useful interpretations for conser- high rate of deforestation (Green & Sussman, 1990; vation. Achard et al., 2002). For species tolerant of habitat disturbance the calculation of PFD using this method would produce an overestimate of population reduction, Methods and therefore we have evaluated such species based on Geographic data were compiled from .550 herbarium direct field observations, indicating those known to be in specimens representing the 91 endemic species of continuing decline (without attempting to calculate PFD) Malagasy Pandanaceae currently recognized (Fig. 1). so that the other Red List criteria can be applied. Geographical coordinates for recent collections were The number of subpopulations within the total recorded with a global positioning system or assigned population of a species was estimated by overlaying a post facto for older specimens using a gazetteer of 10 * 10 km grid onto the mapped locality records and botanical collecting localities (Schatz & Lescot, 1996). counting as distinct subpopulations the number of non- All Malagasy Pandanaceae were assessed using the contiguous occupied cells or cell clusters (Schatz et al., IUCN Red List criteria (2001). Threatened species are 2000; Good et al., 2006). The number of subpopulations those evaluated as Critically Endangered (CR), located within protected areas was determined by Endangered (EN), or Vulnerable (VU) with respect to overlaying the distribution of subpopulations onto a extinction risk in the wild. We have argued elsewhere map of protected areas. (Callmander et al., 2005) that the data deficient (DD) The grid cell size used to calculate AOO influences category should be used with parsimony to avoid Red List assessments (Willis et al., 2003) and because

ß 2007 FFI, Oryx, 41(2), 168–176

Fig. 1 All collections of the genera Martellidendron and Pandanus (Pandanaceae) in Madagascar mapped on the bioclimatic zones of Cornet (1974) simplified to five categories (D, Dry; H, Humid; M, Montane; SA, Subarid; SH, Subhumid; Schatz, 2000), and the main existing Protected Areas (vertical hatching).

ß 2007 FFI, Oryx, 41(2), 168–176

Malagasy Pandanaceae exhibit a wide range of distribu- reciprocal range size for all species per grid cell), species tion types they are an appropriate test group for richness (a count of species per grid cell), irreplaceability evaluating the impact of using alternative grid cell (presence in a cell of the only known populations of one sizes. The Red List guidelines (IUCN & SSC, 2004) or more species), and rarity (the summed inverse range indicate that when selecting a grid ‘the appropriate scale size for all species present in a cell). The latter is a [for calculating AOO] depends on the taxon in question function of both endemism and richness, and is also and the comprehensiveness of the distribution data’, referred to as endemism richness (Kier & Barthlott, with a 2 * 2 km cell size suggested for most situations. 2001). WORLDMAP was also used to calculate the Gardenfors (2001) recommended using smaller cells if a correlation between richness and rarity using species is suspected of being CR but this presupposes an Spearman’s non-parametric rank correlation coefficient IUCN assessment (Willis et al., 2003). Schatz et al. (2000) rho. Analyses were performed on the 87 species for used 10 * 10 km cells to analyse species in Madagascar’s which adequate information is available; the 4 species endemic plant families, a size considered to correspond assessed as DD (Appendix) were excluded. Geographic to the average extent of an isolated subpopulation (Good distribution was also plotted on maps of remaining et al., 2006). Willis et al. (2003) proposed using grid cells vegetation (Steiniger et al., 2003) and elevation (US equal to 1/10 of the length between the most distant pair Geological Survey, 1996). of points of the EOO polygon. While this works well for approximately circular to oval distributions (length to Results width ratio (3:1), it is problematic for taxa with long, narrow distributions, yielding very large cells. We A summary of the number of species assigned to each calculated estimates of AOO using three methods: that Red List category using the three methods of assessment of (1) Willis et al. (2003), (2) Schatz et al. (2000), and (3) a is presented in Table 1. Application of the three methods 3 * 3 km grid. The latter corresponds closely to the near to data from herbarium specimens yielded striking maximum for a taxon to be listed as CR under Red List differences in the proportion of species assessed as criterion B2 (for which the threshold is 10 km2), and better threatened, from 64% using the approach of Willis et al. reflects the overall poor level of geographic sampling of (2003) to 100% using a 3 * 3 km grid cell size. Using a 3 * plants in Madagascar than a 2 * 2 km cell size. 3 km grid but taking field observations of population In calculating EOO ‘disjunctions or discontinuities decline and sensitivity to disturbance into account within the overall range of a taxon’ may be excluded resulted in 95% of the species assessed as threatened (IUCN & SSC, 2004), an idea explored but ultimately (Table 1). The Appendix gives the Red List assessments rejected by Willis et al. (2003) because an element of of all 91 species assessed using this latter method. All uncertainty is introduced. We tested the effect of nine CR species belong to Pandanus, only one of which removing clearly unsuitable habitat (sea and areas (P. insuetus) occurs in a protected area. Four of the eight .800 m elevation) on estimates of EOO using data for CR species not encompassed within protected areas Martellidendron, all of which are restricted to lowland occur in littoral forest, and one each in lowland humid habitats and threatened by fragmentation and forest forest, mid elevation forest, remnant forest parcels on loss. the Central High Plateau, and western/northern decid- Geographic data on Malagasy Pandanaceae were uous forests. Among the 74 species categorized as EN or analysed with WORLDMAP (Williams, 2000) using J- VU, 12 are not known to occur in protected areas and degree grid cells (c. 27 * 27 km) to calculate range nine of these are endemic to littoral forest, two species size rarity (expressed as the percentage aggregated are restricted to low and mid elevation forest, and one to

Table 1 Number of Malagasy Pandanaceae (and % of 87 species assessed) assigned to Red List categories (IUCN, 2001) based on application of the methodologies proposed by Schatz et al. (2000), Willis et al. (2003), a 3 * 3 km grid cell size, and a 3 * 3 km grid incorporating field observations (see text for details). The four species categorized as Data Deficient are omitted.

IUCN category* Total no. threatened Method CR EN VU LC (CR+EN+VU)

Willis et al. (2003) 10 (11) 21 (24) 25 (29) 31 (36) 56 (64) Schatz et al. (2000) 6 (7) 41 (47) 38 (44) 2 (2) 85 (98) 3 * 3 km grid 9 (10) 62 (71) 16 (18) 0 (0) 87 (100) 3 * 3 km grid with field observations 9 (10) 31 (36) 43 (49) 4 (4) 83 (95)

*CR, Critically Endangered; EN, Endangered; VU, Vulnerable; LC, Least Concern

ß 2007 FFI, Oryx, 41(2), 168–176

the central highlands. The four species (4%) assessed as inappropriate for taxa with widespread distributions Least Concern (which indicates they are not threatened that span both suitable and intervening unsuitable with extinction according to the Red List criteria) are habitat. This method is also inappropriate for linear or widespread (EOO .20,000 km2). Four species (4%) were highly disjunct distributions because it leads to the categorized as DD because their taxonomic status has selection of large grid cells, resulting in an under- not yet been fully clarified. assessment of threatened species (Table 1). Furthermore, When calculations of EOO were adjusted by removing this method incorporates a dependency of AOO on clearly unsuitable habitat for the two species of EOO, exacerbating an already inherent scale depen- Martellidendron assessed as EN, the estimated EOO values dency. In these instances the method proposed by were reduced from 1,960 to 863 km2 for M. androcepha- Schatz et al. (2000), using 10 * 10 km grid cells to lanthos and from 717 to 389 km2 for M. karaka (Fig. 2C). calculate AOO, is preferable in Madagascar (Good et al., The highest local species richness of Malagasy 2006); e.g. for littoral forest endemics restricted to a Pandanaceae (10 spp.) occurs in a single J-degree grid narrow band (,10 km wide) along the east coast cell containing the Manongarivo Special Reserve in the (Consiglio et al., 2006). This method is, however, less north-east of the island (Fig. 1). Two other protected well suited for species with very restricted ranges areas (Tsaratanana and Perinet) each contain nine because AOO is overestimated, leading again to an species. Other areas of high species richness (7–8 spp.) under-assessment of CR species (Table 1). A combined include two National Parks within humid forest approach using one or other of these methods, depend- (Mananara-Nord and Marojejy) and several unprotected ing on distribution shape and fragmentation, would littoral forest sites along the east coast (Fig. 2A). exploit their respective strengths, although choosing Centres of high endemism, as indicated by the highest between them is necessarily subjective, introducing a endemism richness scores (7.66–14.35%) are found in bias that could, however, be avoided by consistently four protected areas (Tsaratanana, Manongarivo, employing a single grid size, reducing uncertainty Marojejy and Montagne d’Ambre; Fig. 2A). Relatively (Keith et al., 2000). For Malagasy Pandanaceae use of a high rarity values (.3%) also occur at Perinet and in 3 * 3 km grid results in the highest number of species several unprotected littoral forest sites. There was a assessed as threatened (Table 1) of the three methods strong positive correlation (rho 5 0.782) between used, and may not best reflect biological reality. A more endemism richness and species richness. realistic approach is to incorporate field observations Nineteen species (21%) do not occur in any protected (Appendix, Table 1) that evaluate population decline, area (Fig. 2B) and a minimum of 10 cells are necessary to although it remains impossible to eliminate assessor bias encompass them all, four of which are irreplaceable. altogether (Regan et al., 2005). In reality, complete These same 10 cells collectively encompass 48% (44) of objectivity is unachievable; selecting a grid size for all species. The minimum number of cells required to uniform application, as recommended by IUCN, is itself ensure that all Malagasy Pandanaceae are represented at subjective. least once includes 16 additional cells (i.e. 26 in total). Estimates of EOO can be improved by removing All 87 Malagasy Pandanaceae (excluding DD species) unsuitable habitat (IUCN, 2001) or by using an alpha- occur within the cells collectively occupied by the 28 hull, in which a highly disjunct distribution is separated rarest species (i.e. those occurring in only one or two into several distinct hulls (Burgman & Fox, 2003). Such cells). methods can, however, be difficult to apply using collection data and may lead to substantial biases if there are few records or if recorded locations are Discussion imprecise or contain errors. While no change in threat Pandanaceae are important components of many vege- classification resulted when we removed unsuitable tation types in Madagascar, although they are best habitat for species of Martellidendron, the different EOO represented in humid and subhumid forests and values obtained illustrate how such corrections could riparian communities. They also present a wide variety significantly affect Red List assessments. of distribution patterns, including widespread to locally The highest levels of both species richness and restricted taxa, species with nearly circular or linear endemism richness occur in northern Madagascar, ranges, and with almost continuous to highly disjunct especially in the higher mountains (Manongarivo, and fragmented distributions. Although the method of Marojejy and Tsaratanana), where 95% of the species Willis et al. (2003) is useful for calculating AOO for growing at .800 m altitude are local endemics (i.e. species with restricted ranges and approximately circu- restricted to one or few cells). These results are not lar distributions, setting grid cell size using 1/10 of surprising given that nearly one third of Malagasy the maximum distance between subpopulations is Pandanaceae (27 of 91 spp.) are endemic to the northern

ß 2007 FFI, Oryx, 41(2), 168–176

(A) (B)

(C)

Fig. 2 (A) Endemism richness (see text for details) for all 87 species of Pandanaceae assessed (i.e. excluding the 4 species assessed as Data Deficient) and the four Protected Areas (vertical hatching) with the highest scores. (B) Endemism richness of 19 species not currently included in Madagascar’s protected area network. (C) Distribution of two Endangered species of Martellidendron in northern Madagascar, with remaining forests as of 2003 (map from Steiniger et al., 2003).

ß 2007 FFI, Oryx, 41(2), 168–176

Table 2 Priority areas for Pandanaceae conservation, with size of area and threatened species of Pandanaceae occurring in each location.

Location Area (ha) Species (Red List category)*

Littoral forests Baie d’Antongil (forests around the bay) 185 Martellidendron karaka (EN), Pandanus bipyramidatus (VU), P. columnaris (VU), P. concretus (VU), P. guillaumetii (VU), P. insuetus (CR), P. imerinensis (VU), P. longecuspidatus (VU), P. princeps (CR) Analalava-Foulepointe-Soanierana-Ivongo-Tampolo 2,613 P. callmanderiana (CR), P. cephalothus (VU), P. comatus (EN), P. concretus (VU), P. guillaumetii (VU), P. malgassicus (VU), P. neoleptopodus (EN), P. pseudocollinus (EN), P. platyphyllus (EN), P. rollotii (VU), P. sylvicola (CR) Fort-Dauphin region (Saint-Luce) 457 P. bipyramidatus (VU), P. dauphinensis (CR), P. longistylus (EN), P. peyriarasii (CR), P. platyphyllus (VU), P. rollotii (VU) Natural vegetation of the highlands Andringitra region (E side outside the protected area) 46,649 P. ambalavaoensis (CR) Itremo region (forest patches) 405 P. bakeri (CR) Northern and Western deciduous forests Cap Saint Andre´ (40 km south) 5,858 P. ambongensis (VU), P. perrieri (CR), P. stellatus (VU) Forest block in the Daraina region (Ambilondamba, 17,796 P. analamerensis (VU), P. coriaceus (VU), Ampondrabe, Antsahabe, Binara & Bobankora) P. grallatus (EN), P. latistigmaticus (EN), P. perrieri (CR) Mangoky basin (forest blocks along rivers in the Mangoky Basin) 84,477 P. ambongensis (VU) Eastern and Western rainforests Lakato forest 555 M. cruciatum (VU), P. freycinetioides (VU) Galoko massif (NE of Ambanja) 8,720 M. androcephalanthos (EN), M. cruciatum (VU), P. sp. nov ‘neosermollii’ ined. (CR), P. mammillaris (EN) South of Moramanga (forest patches around Anosibe an’ala) 1,730 P. ceratophorus (VU), P. kimlangii (VU), P. leptopodus (VU), P. macrophyllus (CR)

*Species not occurring in the protected area network are in bold. CR, Critically Endangered; EN, Endangered; VU, Vulnerable

part of the island (Callmander et al., 2003c). However, most considerably less today. Neither of these vegetation threatened Pandanaceae occur in low to mid elevation types has adequate coverage in protected areas, even eastern humid forests, including littoral forests (Schatz though they contain large numbers of locally endemic et al., 2000; Randrianasolo et al., 2002; Good et al., 2006). species. Littoral forests are severely under-represented Our analyses show that 19 Malagasy Pandanaceae are in the protected area network, with just four small not included in the country’s protected area network but parcels totalling 695 km2 (ANGAP, 2005; Consiglio et al., can be encompassed in a near minimum set of 10 J- 2006). Of the 10 new sites, 3 have been granted degree cells, and these therefore represent the highest preliminary protected area status (Daraina, Analalava- priorities for Pandanaceae conservation. If these cells are Tampolo, and Saint Luce). added to the existing protected area network, which One goal of conservation planners is maximum already includes the 16 additional cells required to inclusion of biodiversity in protected areas. Our study encompass all Malagasy Pandanaceae, each member of shows that by conserving sites in Madagascar with the the family would be protected in at least one site. These rarest Pandanaceae (those occurring in only one or a few 10 new priority areas include four major vegetation areas) along with species not known to occur in types (Table 2): littoral forests, natural vegetation on the protected areas, we can achieve protection for the entire central highlands, deciduous forests in the north and family. This is also true for Madagascar’s endemic plant west, and rainforests in the east and north-west. The families (Schatz et al., 2003; Raharimampionona et al., first two are suffering exceptionally high rates of 2006), some members of which are restricted to these deforestation; in 1990 they had ,10% of their original same areas. Application of this method to other groups cover remaining (Green & Sussman, 1990) and have (now being conducted) is confirming the importance of

ß 2007 FFI, Oryx, 41(2), 168–176

these sites and revealing additional conservation prio- References

rities. Ultimately, the identification of such key sites is a Achard, F., Eva, H.D., Stibig, H.-J., Mayaux, P., Gallego, J., fundamental component of developing a comprehensive Richards, T. & Malingreau, J.-P. (2002) Determination of strategy for biodiversity protection, which can be deforestation rates in the world’s humid tropical forests. facilitated by carefully evaluating the conservation Science, 297, 999–1002. status of well-delimited species using geo-referenced ANGAP (2005) Liste des Parcs et Re´serves de Madagascar. Http:// www.parcs-madagascar.com/liste.htm [accessed 10 January data from natural history collections. 2007]. Our study confirms that conservation planning in Burgman, M.A. & Fox, J.C. (2003) Bias in species range Madagascar must include as much of the remaining estimates from minimum convex polygons: implications for lowland forest of the east coast as possible, especially conservation and options for improved planning. Animal littoral forests (Consiglio et al., 2006); stands between the Conservation, 6, 19–28. Callmander, M.W., Chassot, P., Ku¨ pfer, P. & Lowry II, P.P. (2003a) Baie d’Antongil (central north-east) and Tampolo Recognition of Martellidendron, a new genus of Pandanaceae, (central east) are of particular importance for and its biogeographic implications. Taxon, 52, 747–762. Pandanaceae, as they are for the endemic plant families Callmander, M.W. & Laivao, M.O. (2002) Re´vision de Pandanus (Schatz et al., 2003). sect. Dauphinensia St. John (Pandanaceae) a` Madagascar. For Red List assessments the appropriate selection of Botanica Helvetica, 112, 47–67. Callmander, M.W., Laivao, M.O. & Wohlhauser, S. (2003b) Les grid cell size is dependent upon the shape, size and Pandanus sect. Acanthostyla Martelli (Pandanaceae) d’altitude homogeneity of a species’ distribution, as manifested in du Nord de Madagascar, avec description de deux nouvelles part by its extent of occurrence. By selectively applying espe`ces. Candollea, 58, 63–74. complementary methods for choosing grid size (Schatz Callmander, M.W., Schatz, G.E. & Lowry II, P.P. (2005) IUCN Red et al., 2000; Willis et al., 2003; IUCN & SSC, 2004), List assessment and the Global Strategy for Plant Conservation: taxonomists must act now. Taxon, 54, 1047–1050. depending on distribution attributes, we have shown Callmander, M.W., Wohlhauser, S. & Gautier, L. (2003c) Notes that there is no single best approach to selecting an bioge´ographiques sur les Pandanaceae du nord de appropriate scale for calculating area of occurrence, and Madagascar. Candollea, 58, 351–367. have demonstrated that the method chosen can influ- Callmander, M.W., Wohlhauser, S. & Laivao, M.O. (2001) Une ence threat assessments. With 75% of Malagasy nouvelle section du genre Pandanus (Pandanaceae) a` ´ Pandanaceae threatened, conservation action must be Madagascar: Pandanus sect. Tridentistigma. Adansonia, ser. 3, 23, 49–57. taken immediately to ensure their continued survival. Consiglio, T., Schatz, G.E., McPherson, G., Lowry II, P.P., The careful analysis of specimen data using appropriate Rabenantoandro, J., Rogers, Z.S., Rabevohitra, R. & methods can make an important contribution to this Rabehevitra, D. (2006) Deforestation and plant diversity of process by identifying threatened species and revealing Madagascar’s littoral forests. Conservation Biology, 20, the highest priority sites for their conservation. 1799–1803. Cornet, A. (1974) Essai de cartographie bioclimatique a` Madagascar. Notice explicative N˚55, ORSTOM, Paris, France. Ga¨rdenfors, U., Hilton-Taylor, C., Mace, G. & Rodrı´guez, J.P. Acknowledgements (2001) The application of IUCN Red List Criteria at regional levels. Conservation Biology, 15, 1206–1212. We thank the Parc Botanique et Zoologique de Good, T.C., Zjhra, M.L. & Kremen, C. (2006) Addressing data Tsimbazaza, ANGAP (Association Nationale pour la deficiency in classifying extinction risk: a case study of a Gestion des Aires Prote´ge´es), and the Missouri Botanical radiation of Bignoniaceae from Madagascar. Conservation Garden’s office in Antananarivo (especially Lalao Biology, 20, 1099–1110. Goodman, S.M. & Benstead, J.P. (2005) Updated estimates of Andriamahefarivo) for local assistance, and Bette biotic diversity and endemism for Madagascar. Oryx, 39, 73–77. Loiselle (University of Missouri–St. Louis, USA), Prof. Green, G.M. & Sussman, R.W. (1990) Deforestation history of Philippe Ku¨ pfer (Universite´ de Neuchaˆtel) and Wendy the eastern rainforest of Madagascar from satellite image. Strahm (IUCN, Gland, Switzerland) for support before Science, 248, 212–215. and during this study. We are grateful to Craig Hilton- IUCN (2001) 2001 Categories and Criteria (version 3.1). IUCN, Gland, Switzerland [http://www.redlist.org/info/ Taylor (IUCN, Cambridge, UK), Mark Burgman categories_criteria2001.html, accessed 20 February 2007]. (University of Melbourne, Australia) and Justin Moat IUCN (2006) 2006 IUCN Red List of Threatened Species. IUCN, (Royal Botanic Gardens, Kew, UK) for kind responses to Gland, Switzerland [http://www.redlist.org, accessed 10 queries regarding methods, and to two anonymous January 2007]. reviewers for helpful comments. This work was sup- IUCN & SSC (2004) Guidelines for using the IUCN Red List ported by grants to MWC from the Swiss National Categories and Criteria. IUCN, Gland, Switzerland. Keith, D.A., Auld, T.D., Ooi, M.K.J. & Mackenzie, D.E. (2000) Science Foundation (No. PBNE2-102378), National Sensitivity analyses of decision rules in World Conservation Geographic Society (No. 04065) and Conservation Union (IUCN) Red List Criteria using Australian plants. International-Madagascar (Convention 234). Biological Conservation, 94, 311–319.

ß 2007 FFI, Oryx, 41(2), 168–176

Kier, G. & Barthlott, W. (2001) Measuring and mapping www.mobot.org/MOBOT/Madagasc/Rediscovering% endemism and species richness: a new methodological 20Lost%20Species_files/frame.htm [accessed 10 January 2007]. approach and its application on the flora of Africa. Schatz, G.E., Birkinshaw, C., Lowry II, P.P., Randriantafika, F. & Biodiversity and Conservation, 10, 1513–1529. Ratovoson, F. (2000) The endemic plant families of Madagascar Laivao, M.O., Callmander, M.W. & Buerki, S. (2006) Sur les project: integrating taxonomy and conservation. In Diversite´ et Pandanus (Pandanaceae) a` stigmates saillants de la coˆte est de Ende´misme a` Madagascar (eds W.R. Lourenço & S.M. Goodman), Madagascar. Adansonia, se´r. 3, 28, 267–285. pp. 11–24. Me´moires de la Socie´te´ Biogeógraphie, Paris, France. Laivao, M.O., Callmander, M.W. & Wohlhauser, S. (2000) Une Schatz, G.E. & Lescot, M. (1996) Gazetteer to Malagasy Botanical espe`ce nouvelle de Pandanus sect. Martellidendron Collecting Localities. Http://www.mobot.org/MOBOT/ (Pandanaceae) de la peńinsule de Masoala, Madagascar. research/madagascar/gazetteer/ [accessed 10 January 2007]. Botanica Helvetica, 110, 41–49. Steiniger, M.K., Harper, G.J. & Tucker, C.J. (2003) Forest Cover Myers, N. & Knoll, H.A. (2001) The biotic crisis and the future Fragmentation and Clearance in Madagascar. Centre for Applied of evolution. Proceedings of the National Academy of Sciences, 98, Biodiversity Science, Conservation International, Arlington, 5389–5392. USA & Nasa Goddard Space Flight Center, Washington, DC, Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, USA. G.A.B. & Kent, J. (2000) Biodiversity hotspots for US Geological Survey (1996) Global 30-Arc-Second Elevation Data conservation priorities. Nature, 403, 853–858. Set. US Geological Survey, Sioux Falls, USA. Phillipson, P.B., Schatz, G.E., Lowry II, P.P. & Labat, J.-N. (2006) Williams, P.H. (2000) WORLDMAP. Software and help document A catalogue of the vascular plants of Madagascar. In 4.2. Privately distributed, London, UK. Taxonomy and Ecology of African Plants: Their Conservation Willis, F., Moat, J. & Paton, A. (2003) Defining a role for and Sustainable Use (eds S.A. Ghazanfar & H.J. Beentje), pp. herbarium data in Red List assessments: a case study of 613–627. Proceedings 17th AETFAT Congress, Addis Ababa, Plectranthus from eastern and southern tropical Africa. Ethiopia & Royal Botanic Gardens, Kew, London, UK. Biodiversity and Conservation, 12, 1537–1552. Raharimampionona, J., Andriambololonera, S., Schatz, G.E., Lowry II, P.P., Rabarimanarivo, M., Ratodisoa, A. & Appendix Ravololomanana, N. (2006) Identification des aires prioritaires pour la conservation des plantes a` Madagascar: The Appendix for this article is available online at utilisation des donneés botaniques pour de´finir les priorite´s http://journals.cambridge.org en matie`re de conservation. In Taxonomy and Ecology of African Plants: Their Conservation and Sustainable Use (eds S.A. Ghazanfar & H.J. Beentje), pp. 447–456. Proceedings 17th Biogeographical sketches AETFAT Congress, Addis Ababa, Ethiopia & Royal Botanic Gardens, Kew, London, UK. Martin W. Callmander’s research interests include the Randrianasolo, A., Miller, J.S. & Consiglio, T.K. (2002) inventory and systematics of the Malagasy flora, Malagasy Application of IUCN Criteria and Red List Categories plant biogeography, and using botanical data to evaluate on species of five Anacardiaceae genera in Madagascar. endemism and conserve the remaining forests of this Biodiversity and Conservation, 7, 1289–1300. biodiversity hotspot. Regan, T.J., Burgman, M.A., McCarthy, M.A., Master, L.L., Keith, D.A., Mace, G.M. & Andelman, S.J. (2005) The George E. Schatz’s research interests include the woody flora of consistency of extinction risk classification protocols. Madagascar, the use of primary occurrence data for extinction Conservation Biology, 19, 1969–1977. risk assessment, and species distribution modelling. Rodrigues, A.S.L., Pilgrim, J.D., Lamoreux, J.F., Hoffman, M. & Porter P. Lowry II’s research interests include the inventory Brooks, T.M. (2006) The value of the IUCN Red List for and systematics of the Malagasy flora, and assessing conservation. Trends in Ecology & Evolution, 21, 71–76. patterns of biological diversity to identify threatened species Schatz, G.E. (2000) Endemism in the Malagasy tree flora. In and priority areas for conservation. Diversite´ et Ende´misme a` Madagascar (eds W.R. Lourenço & S.M. Goodman), pp. 1–9. Me´moires de la Socie´te´ Michel O. Laivao, J. Raharimampionona, S. Andriam- Biogeógraphie, Paris, France. bololonera and T. Raminosoa work for Missouri Botanical Schatz, G.E. (2001) Generic Tree Flora of Madagascar. [Flore Geńe´rique Garden’s research and conservation programme in des Arbres de Madagascar]. Missouri Botanical Garden, Saint- Madagascar. They are currently working on a threat Louis, USA & Royal Botanic Gardens, Kew, UK. assessment of endemic Malagasy plant species. Schatz, G.E. (2002) Taxonomy and herbaria in service of plant Trisha Consiglio specializes in geographical information conservation: lessons from Madagascar’s endemic families. systems, using tools that include remote sensing and species Annals Missouri Botanical Garden, 89, 145–152. distribution modelling to identify key areas of plant Schatz,G.E.,Andriambololonera,S.,Randriampionona,J.&Lowry diversity and endemism. II, P.P. (2003) Rediscovering Lost Species. Http://

ß 2007 FFI, Oryx, 41(2), 168–176